Epigenetic Inactivation of Acetyl-Co A acetyltransferase 1 promotes the proliferation and metastasis of nasopharyngeal carcinoma via decreasing ketogenesis

Yunliang Lu Guangxi Medical University Xiaohui Zhou Guangxi Medical University Weilin Zhao Guangxi Medical University Zhipeng Liao Guangxi Medical University Bo Li Guangxi Medical University Yanping Yang Guangxi Medical University Xuemin Zhong Guangxi Medical University Yingxi Mo Guangxi Medical University Ping Li Guangxi Medical University Guangwu Huang Guangxi Medical University Xue Xiao Guangxi Medical University Zhe Zhang Guangxi Medical University Xiaoying Zhou (  zhouxiaoying1982@outlook.com ) Guangxi Medical University


Abstract
Background Acy1 Coenzyme A Acyltransferases1 (ACAT1) is a key enzyme in the metabolism of ketone bodies, but its expression and biological function in the pathogenesis of NPC remains underexplored.

Methods
The mRNA and protein expression levels of ACAT1 in NPC and normal control tissues were analyzed by qPCR and immunohistochemistry staining, respectively. GEO database was applied for meta-analysis of ACAT1 mRNA expression and DNA promoter methylation. The role of ACAT1 in NPC proliferation was examined by CCK8 and colony formation assays in vitro and tumorigenicity in vivo. The wound healing and transwell assays were used for analyzing the migratory and invasive ability. cDNA microarray analysis was performed to identify the genes involved in epithelial-mesenchymal transition and dysregulated by ACAT1. These changes were further con rmed by western blot.

Results
We found that ACAT1 is inactivated in NPC cell lines and primary tissues. DNA microarray data showed higher methylation in the CpG island region of ACAT1 in NPC than normal tissues. The demethylating reagent 5-aza-dC signi cantly restored the transcription of ACAT1 in NPC cell lines, suggesting that ACAT1 was inactivated by DNA promoter hypermethylation. Ectopic overexpression of ACAT1 remarkably suppressed the proliferation and colony formation of NPC cells in vitro. As well, the tumorigenesis of NPC cells overexpressing ACAT1 was decreased in vivo. In addition, the migratory and invasive capacities of NPC cells was inhibited by ACAT1 overexpression. Importantly, the higher level of ACAT1 was accompanied by an increased expression of CDH1, EPCAM, and a decreased expression of vimentin and SPARC. This strongly indicates that ACAT1 is able to affect the epithelial-mesenchymal transition in NPC, thereby controlling cellular motility. In addition, we found that ACAT1 expression increases the intracellular level of β-HB. Moreover, exogenous β-HB remarkably inhibits the growth of NPC cells in a dose-dependent manner.

Conclusions
We have discovered that the ketone body metabolism enzyme ACAT1 is epigenetically downregulated in NPC and acts as a potential tumor suppressor in NPC. Our ndings highlight the possibility of using the modulation of ketone body metabolism as effective adjuvant therapy for NPC.

Page 4/23
Aberrant energy metabolism was de ned as one of the hallmarks of cancer (1). The metabolic reprogramming in cancer includes the generation of additional energy, providing more substrates for biosynthesis, and rebalancing cellular redox status (2), thus facilitating the adaption of tumor cells to the tumor microenvironment and generating extra energy to sustain their malignant growth (3)(4)(5)(6).
Nasopharyngeal carcinoma (NPC) derives from the tissue of the nasopharynx. It is the most common one of the head and neck cancer prevalent in southern China and Southeast Asia (7). Synergistic effects of genetic susceptibility, Epstein-Barr virus (EBV) infection, and environmental carcinogens are considered to be the main etiologic factors of NPC (8). More than 75% of NPC patients were at an advanced stage when diagnosed in the clinic (9). The intensity-modulated radiotherapy (IMRT) has improved the disease control of NPC, but its effect on local recurrence and distant metastasis is limited (10). Therefore, further insights into the carcinogenesis of NPC are necessary to develop new therapeutic approaches.
In the previous study, we found that lipid droplets (LDs) were accumulated in NPC, and the amount of intracellular LDs is associated with the growth of NPC (11), indicating the existence of aberrant lipid metabolism. As raw material for lipid synthesis, Acetyl-CoA is produced during ketone body metabolism, which is a central node in physiological homeostasis. Recently, we reported that a ketone body biosynthesis enzyme, 3-hydroxybutyrate dehydrogenase type 2 (BDH2), was signi cantly inactivated in NPC (12). In addition, an essential rate-limiting enzyme in ketogenesis, hydroxymethylglutaryl-CoA lyase (HMGCL), was found to be downregulated in NPC. Overexpression of HMGCL restores the production of β-hydroxybutyrate (β-HB), the main component of ketone bodies, thereby impeding NPC cell proliferation and metastasis (13). Our data highlights a remarkably reprogrammed ketone body metabolism in NPC, which warrants further investigation. The enzymes responsible for this might be considered as novel candidate targets for NPC therapy.
Acetyl-CoA acetyltransferase 1 (ACAT1) is located mitochondria and catalyzes the reversible formation of acetoacetyl-CoA from two molecules of acetyl-CoA, as the rst step of ketogenesis. To date, most studies have focused on mutations of ACAT1 leading to β-ketothiolase de ciency (14). Publications about the dysregulation of ACAT1 and its role in tumor pathogenesis are rare. The expression of ACAT1 was decreased in clear cell renal cell carcinoma (ccRCC) (15). Restoring its expression signi cantly reduces the malignant behavior of ccRCC cells in vitro (16), suggesting that ACAT1 might be a tumor suppressor. On the contrary, as an effective diagnostic and prognostic biomarker, ACAT1 was elevated in aggressive prostate cancer (17,18). However, the role of ACAT1 in NPC remains unclear.
In this study, we investigated the expression of ACAT1 in NPC, bring to light the role of ACAT1 in the biological behavior of NPC and the underlying molecular mechanisms relating to ketone body metabolism.
In all, primary NPC tumor tissues were obtained from 42 diagnosed and untreated cases with informed consent from the donors in the Department of Otolaryngology-Head and Neck Surgery, First A liated Hospital of Guangxi Medical University (Nanning, China). The diagnoses were established by experienced pathologists according to the World Health Organization (WHO) classi cation. We also included 36 normal nasopharyngeal epithelial tissues as control.
The demethylating reagent 5-aza-dC was purchased from Sigma (#A3656, USA). Full-length cDNA from the open reading frame of ACAT1 (Origene, USA) was subcloned into the pCMV6-Entry vector (Origene, USA). NPC cells were cultured in 6-well dishes to 70-90% con uence then transfected with 2μg pCMV6-Entry or ACAT1 plasmid using an X-treme GENE HP DNA Transfection Reagent (Roche, Germany) for 48 hrs.

Immunohistochemical staining
After depara nization and rehydration, the antigen was retrieved in 5% urea buffer by microwave heating for 5 min, and then incubated in 3% H 2 O 2 for 30 min to block endogenous peroxidase activity. Tissue sections were incubated with the rst antibody overnight at 4℃ and followed by a secondary antibody for 1 h at room temperature. Subsequently, 3,3'-diaminobenzidine (DAB) reagent (ZLI-9018, ZSGB-BIO, Beijing) was used for peroxidase reaction and hematoxylin was used for counterstaining. Images were acquired under a microscope (Olympus C-5050, Japan).

In vivo tumorigenesis
The animal study was approved by the Animal Ethical Committee of First A liated Hospital of Guangxi Medical University. All the methods were carried out in accordance with the approved guidelines. Five 5week-old male BALB/c-nu nude mice (Vital River Laboratory Animal Technology, China) were injected with 1×10 6 pCMV6-Entry-5-8F/HONE1 cells in the right ank and an equal amount of ACAT1-5-8F/HONE1 cells was injected into the left ank. The tumor volume was assessed by 2D measurements at 3, 6, 9, 12, 15, 18 days. Tumor volume was calculated as volume (mm 3 ) = length×width×0.5. Eighteen days for 5-8F and fteen days for HONE1 after inoculation, all mice were killed and tumors were removed. cDNA microarray analysis Affymetrix GeneChip Genome U133 Plus 2.0 expression array and Gene-Cloud of Biotechnology Information (GCBI) platform (www.gcbi.com.cn) were used to analyze genes with differential expression between ACAT1-5-8F and pCMV6-Entry-5-8F cells. The microarray data is available via the following accession identi er on the NCBI-GEO database: GSE155206.
Wound healing Cells (8×10 5 per well) were seeded into 12-well plates in DMEM medium without FBS overnight. Monolayer cells were scratched by an ibidi Culture-Insert (No. 80209, ibidi, Germany). Images were acquired under an inverted phase microscope (TS100, Nikon, Japan) at 0 h and 24 hrs. The migratory distance was analyzed by Image J ver.1.51k (NIH, USA) software.

Transwell
Cells (7×10 4 per well) suspended in 200 μl of serum-free medium were seeded in the upper transwell chamber (8 mm pores, Corning, USA), pre-coated with matrigel (BD Biosciences). The lower chamber was lled with DMEM medium with 10% FBS. Non-invading cells were removed by using a cotton-tipped swab after 36 hrs. Invasive cells on the lower membrane surface were xed with 1% paraformaldehyde, stained with 0.5% crystal violet, and photographed.

β-hydroxybutyrate (β-HB) detection
Cells were grown in 60-mm dishes in serum-free DMEM medium for 48 hrs and then lysed with RIPA buffer. β-HB colorimetric assay kit (#700190, Cayman Chemical, Ann Arbor, MI, USA) was used for determining the intracellular β-HB level as previously (19).

Statistical analysis
All data were analyzed using SPSS 20.0 (SPSS Inc., Chicago, IL, USA). Data are expressed as mean±SD and were analyzed by Pearson's chi-square test and Fisher's exact test. Statistical signi cance was considered at *p< 0.05, **p< 0.01 and ***p< 0.01.

Results
The expression of ACAT1 is signi cantly downregulated in NPC.
To reveal the alteration of ACAT1 in NPC, we rstly investigated the transcriptional level of ACAT1 in seven NPC cell lines (HK1, HONE1, CNE1, 5-8F, 6-10B, and C666-1) compared with an immortalized normal nasopharyngeal epithelial cell line (NP69). Figure 1A shows a reduction of ACAT1 in NPC cell lines. We further analyzed both mRNA and protein expression of ACAT1 in NPC tissues and non-cancer control tissues. We found that the transcription of the ACAT1 gene was signi cantly downregulated in 21 NPC primary tumors, but easily detected in all 23 NNE samples (Fig. 1B). In order to evaluate the consistency of abnormal transcription of ACAT1 in NPC, we performed a meta-analysis using 6 sets of microarrays from the GEO database including 143 NPC tissues and 42 normal tissues (Table S1). The results generated by the random-effects model indicated that signi cant heterogeneity existed among individual datasets (I 2 = 61.2%,p = 0.025) and the pooled Standard Mean Difference (SMD) as -0.98 (95% CI: -1.63,-0.33, Fig. S1A). No signi cant difference was found in the sensitivity analysis (Fig. S1B). The results of Begg's test showed no signi cant publication bias (p = 0.452, Fig. S1C ).
Besides, we also found that the mRNA level of ACAT1 in head and neck squamous carcinoma (HNSC) was decreased in contrast with normal control tissues based on the TCGA database (Fig. S1), which is in line with our ndings in NPC. The lower mRNA expression of ACAT1 was positively correlated with HNSC patients older than 60 years, and higher histologic grade (Table S2).
In addition, we measured the protein levels of ACAT1 in NPC tissue and NNE tissues. ACAT1 protein was localized in the cytoplasm of cells and was highly expressed in the NNE layer (n = 19) (Fig. 2), while almost absent in NPC tissue (n = 15). This nding further supports that ACAT1 is inactivated in NPC. ACAT1 is inactivated by DNA promoter hypermethylation in NPC.
To address the mechanism for the inactivation of ACAT1 in NPC, we rstly analyzed the miRNAs potentially targeting ACAT1, including miR-21, miR-23b, miR-1260, and miR-9, which were screened by the Gene-Cloud of Biotechnology Information platform (Fig. S3A). However, we could not nd miRNAs that were signi cantly overexpressed in primary NPC tissues as compared to normal control samples based on the miRNA microarray database (GSE32960) (Fig. S3B). The expression of miR-21 and miR-1260 is even decreased signi cantly in NPC samples.
Interestingly, we found that a CpG island with a length of 241 bp (-295bp -55 bp from the transcription starting site) was found in the DNA promoter region of ACAT1. We analyzed the methylation microarray dataset (GSE62336) and found that the methylation modi cation in the CpG island region of ACAT1 is remarkably heavier in NPC (n = 25) in contrast to NNE (n = 25) tissues (Fig. 3A). In addition, the mRNA level of ACAT1 was restored signi cantly upon treatment with the demethylation reagent 5-aza-dC in the NPC cell lines 5-8F, HONE1, and TW03. The mRNA level of ACAT1 was elevated in HK1 by 5-aza-dC treatment as well, but no signi cant difference was observed (Fig. 3B). These data suggest that ACAT1 might be inactivated in NPC by DNA promoter hypermethylation.
Furthermore, we found that the level of methylation in the promoter region of the ACAT1 gene is higher also in HNSC, based on the TCGA database (Fig. S4A). Importantly, the transcriptional level of ACAT1 is inversely correlated with the methylation level in its promoter, which strongly supports hypermethylation as the reason for ACAT1 inactivation.

Overexpression of ACAT1 suppresses cell proliferation and colony formation of NPC cell lines in vitro.
To investigate the biological function of ACAT1 in NPC, we established 5-8F and HONE1 cell lines stably overexpressing of ACAT1. Western blot results con rmed that ACAT1 was successfully expressed in both cell lines (Fig. 4A). The proliferative rate of ACAT1-5-8F/HONE1 cells was remarkably slower than the control cells (Fig. 4B). Also, the capacity of colony formation of both NPC cell lines was reduced signi cantly by ACAT1 overexpression (Fig. 4C). Therefore, our nding indicates that overexpression of ACAT1 suppresses the growth of NPC cells in vitro.

ACAT1 suppresses NPC tumor formation in vivo.
To con rm the suppressive effect of ACAT1 expression on NPC tumor cell growth, we examined the tumorigenesis of ACAT1-5-8F/HONE1 and pCMV6-Entry-5-8F/HONE1 in vivo. We observed that the average volume of tumor-derived from ACAT1-5-8F/HONE1 is lower than that from pCMV6-Entry-5-8F/HONE1 (Fig. 5). However, a statistical signi cance was observed only in the 5-8F cell line.

ACAT1 suppresses NPC cell migration and invasion via epithelial-mesenchymal transition.
We further addressed the effect of ACAT1 on migratory and invasive capacities in NPC cells. The wound healing assays revealed a slower gap closure in ACAT1-5-8F and ACAT1-HONE1 cells in contrast to control cell lines (Fig. 6A). In addition, the number of invading ACAT1-5-8F and ACAT1-HONE1 cells was lower than for the control cell lines (Fig. 6B). Our nding suggests that overexpression of ACAT1 attenuates the migration and invasion of NPC cells.
To investigate the potential molecular mechanism by which ACAT1 affects migratory and invasive capacity, we performed cDNA microarray analysis in ACAT1-5-8F and PCMV6-Entry-5-8F cells. We found that genes involved in epithelial-mesenchymal transition (EMT) were signi cantly altered by ACAT1 overexpression, including downregulation of SPARC, ZEB2, etc. and upregulation of EPCAM, CDH1, and CDH3, etc. (Fig. 6C). In addition, we analyzed the expression of E-cadherin and vimentin by western-blot. In comparing with control cell lines, E-cadherin was upregulated in ACAT1-5-8F/HONE1 cells, while vimentin was downregulated (Fig. 6D). Therefore, overexpression of ACAT1 may reverse the EMT process, thereby reducing the metastatic potential of NPC cells.

Discussion
Functioning as a kind of high-energy mitochondrial fuels, ketone bodies are normally generated in hepatocytes and used during starvation (20). Recent studies provide evidence that ketone bodies can be produced in cancer cells as well as in tumor stroma (19,21). In this study, we show that the enzyme ACAT1 catalyzing the rst step of ketogenesis is inactivated in NPC cell lines and primary tumor tissues.
This further supports our previous nding that the production of ketones in NPC is inhibited. On the other hand, ACAT1 is also required for the utilization of ketone bodies. This suggests that NPC cells tend to provide and utilize the energy for promoting growth in other sources than ketones. Thus, a signi cant ketone body metabolism-inactivation pattern is one of the hallmarks of NPC. Besides ACAT1, the other two key enzymes BDH1 and OXT1 were all downregulated in Glioblastoma (22). The expression of ACAT1 is regulated by the well-studied onco-microRNA miR-21, which is elevated in various tumors (23).
We found a CpG island in the promoter region of ACAT1. Based on the methylation microarray data from the GEO database, the methylation level of this CpG island region is signi cantly higher in NPC tissues in contrast to normal control tissues. In addition, 5-aza-dC treatment upregulated the mRNA level of ACAT1, further indicating that ACAT1 is inactivated by DNA hypermethylation in NPC. To our knowledge, this is a novel mechanism for modulating ACAT1 expression in tumors. The upregulation of ACAT1 in breast cancer was shown to induce apoptosis (23). We found that restoring the expression of ACAT1 leads to increased intracellular β-hydroxybutyrate (β-HB) while the proliferation, colony formation, and in vivo tumorigenesis of NPC cells was decreased remarkably. In addition, extracellular β-HB impedes the growth of NPC cells as well. This directly supports the opinion that ketones interfere with NPC cell survival. Previously it was shown that both intracellular and extracellular β-HB induce signi cant generation of ROS, thereby suppressing the viability of NPC (19). As an important factor for tumorigenesis, chronic in ammation promotes tumor development, progression, and metastatic dissemination, as well as treatment resistance. β-HB was identi ed as the factor exerting an anti-in ammatory effect by inhibiting the in ammasome NOD-like receptor pyrin domain-containing 3 (NLRP3) (24,25). In glioma, inhibiting the in ammatory microenvironment suppresses tumor cell migration upon treatment with β-HB, implying that β-HB could be of bene t for glioma therapy (26). In accordance with our study, increasing the β-HB level by overexpressing ACAT1 substantially reduces the motility of NPC cells. 3-Hydroxymethylglutaryl-CoA synthase 2 (HMGCS2) is the enzyme directly downstream of ACAT1 in the ketogenesis pathway (27). Its inactivation in hepatocellular carcinoma has been veri ed and was positively associated with ketone body production. Importantly, the ketone bodies inhibit tumor cell migration by reversing epithelialmesenchymal transition (EMT) signaling (28). Similarly, we showed that the epithelial markers (CDH1 and EPCAM) were increased in NPC cells after overexpressing ACAT1, while mesenchymal markers (vimentin and SPARC) were decreased. Our data indicate that ACAT1 represses the movement of NPC cells by regulating the expression of EMT-related markers.
In order to further uncover the mechanisms underlying ACAT1-mediated control of The EMT signaling, we focused on the expression of CDH1, which is upregulated by overexpressing ACAT1 or by β-HB treatment alone (19). As an endogenous histone deacetylase inhibitor (HDACi), β-HB is able to regulate gene expression affecting several cellular processes (29). We observed increased acetylation of lysine 9 and lysine 14 of histone 3 in the promoter region of CDH1 after the treatment with β-HB, indicating transcriptional stimulation of CDH1 (data not shown). HDACi induced p53-dependent apoptosis in NPC cells (30) and retarded the growth of carcinomas of the cervix, colon, and rectum in vitro (31). Surprisingly, a recent study reported that NPC cells were turned into a mesenchymal cell phenotype after short-term stimulation with the HDACi trichostatin A. However, an invasive phenotype was not induced (32). Additionally, HDACs contribute to maintaining EBV latent infection (33). HDACs, such as TSA, SAHA, and butyric acids, act as potent inducers of EBV reactivation, as well as mediating apoptosis of NPC cells, Therefore, reactivating EBV by HDACi is a potential approach to treat NPC. So far, it remains unknown if EBV reactivation in NPC cell lines can be induced by β-HB. Further veri cation is necessary.
In fact, as non-toxic adjuvant therapy, ketogenic diets or ketone supplementation have shown a positive therapeutic advantage in malignancies. Notably, the preclinical studies demonstrate that a ketogenic diet increases the radiation sensitivity in xenograft models of pancreatic cancer (34). Our previous and present ndings show that overexpressing the ketogenesis genes HMGCL and ACAT1, resulted in increased intracellular β-HB in both renal cell carcinoma and NPC (16,19), thereby inhibiting the growth capacity of tumor cells. As well, exogenous β-HB remarkably suppresses the proliferation and metastasis of NPC cells. Thus, elevating the level of β-HB by targeting ketogenic genes, ketogenic diets or ketone supplementation might be a promising therapeutic approach for NPC patients. This is worthy of further con rmation in animal models.

Conclusions
In summary, we discovered that the expression of ACAT1 was inactivated in NPC due to promoter hypermethylation. Overexpression of ACAT1 elevates the intracellular β-HB in NPC cells, inhibits the proliferation, migration, and invasive growth of NPC cells. Our data reveal that epigenetic modi cation affects the ketone body metabolism which might contribute to pathogenesis of NPC. In addition, our ndings support the manipulation of ketone body metabolism, for example, in a form of a drug or by dietary intervention is a potentially valuable approach for the prevention and treatment of NPC.   Immunohistochemical staining of ACAT1 in NNE tissue (n=19) and NPC (n=15). Magni cations ×200.